Privacy & data ownership

Imports, saved paths and review notes are stored in a server database under your signed-in identity. Reads and writes require authentication and are scoped to the same user. No advertising or third-party analytics is enabled.

Pseudonymous usage measurement

On the public site, page views and aggregate interactions are counted without third-party trackers and without cookies. This is pseudonymous rather than anonymous measurement, and the difference is worth stating precisely: your IP address and User-Agent are processed while your request is handled, and are used to derive a monthly identifier so that return visits can be counted. Neither the raw IP address nor the full User-Agent is ever saved. The identifier is a keyed, salted hash; the salt rotates every month, so a visit this month cannot be linked to a visit next month. Events are bucketed to the hour and deleted after 90 days by a scheduled job. Text you type into comparison or filter inputs is never stored — not even in redacted form; only a hash and a length are kept. Aggregate counts decide which configurations deserve deeper measurement.

Only the documented evaluation fields are accepted. Do not submit credentials, repository contents or raw prompts. Individual saved paths can be removed with the Saved button. Bulk account export/deletion for experiment and review records is a remaining production requirement.